.
Teal overlay

New software edits images based only on users’ thoughts

Many of us are already used to technology you can operate with words or even gestures, but it may soon be possible to interact with software and the devices it powers using only our thoughts.

Researchers from the University of Copenhagen and the University of Helsinki have created a system that is able to edit digital images based entirely on human thoughts.

Associate Professor Tuukka Ruotsalo, from the University of Copenhagen’s Department of Computer Science, explained that the computer is able to make changes with absolutely no prior information about what it should edit or how.

This differs completely from existing AI-powered software that can edit images after extensive training with labelled input, and Prof Ruotsalo said that it was the first time it had been achieved.

The team’s research is being presented at CVPR (Computer Vision and Pattern Recognition) 2022, the most prestigious international conference in this field of research.

Their paper details how 30 participants were given hoods with electrodes to produce a measure of brain activity known as an electroencephalogram or EEG.

Each of the volunteers was given the same set of 200 facial images to look at, along with instructions to look for identifying features such as female faces, older people or people who had blonde hair.

Without performing any other actions, the participants looked at each image for half a second.

The computer would read the EEG to interpret the preference the person was looking for, and then edit the image accordingly.

System can change images to resemble the one that the user is looking for

If the volunteer was looking for older faces, for example, the computer would make younger faces appear older.

If the task was to look for a specified hair colour, the images would be edited so that they all possessed that colour.

PhD student Keith Davis, from the University of Helsinki, said that it was notable that the computer system had no training in facial recognition or knowledge of gender, hair colour and other marked features.

Despite this, it was able to edit only the relevant feature, leaving others untouched, based only on the volunteers’ thoughts as an input.

The researchers called their work “an entirely new paradigm” in AI and said that it could have applications in medicine and for helping people with mobility issues to access computers.

They said that the range of potential applications would be wide and that they also hoped to improve the field of machine learning in general.

Today’s news was brought to you by TD SYNNEX – the UK’s number one solutions distributor.

Back to Top